Simulating the Response of Potato to Applied Nitrogen

Transcription

1 Simulating the Response of Potato to Applied Nitrogen W. Bowen 1, H. Cabrera 2, V. Barrera 3, and G. Baigorria 4 In potato (Solanum tuberosum) production, nitrogen (N) is applied more frequently and in greater amounts than any other nutrient. It is also the nutrient that most often limits yield. Without added N, growing plants often show N deficiency characterized by yellow leaves, stunted growth, and lower yields. Because it is an important input, N and factors affecting its availability have been the subject of much investigation (Harris, 1992). A common objective of many of those studies has been to develop N recommendation systems to assist growers in determining the amount of N needed and the best time to apply it. Knowledge of when and how much N to apply is essential, not only because N inputs have an economic cost but N in excess of that used by the crop may have an environmental cost. The amount of N needed for crop growth and its ultimate fate are important issues regardless of whether the N source is organic matter or mineral fertilizer. An approach often used in field experiments for estimating the amount of N to apply to potato involves measuring yield response to increasing rates of N fertilizer. Yield response is quantified by fitting a mathematical equation to the data. This equation is then used together with economic information to investigate returns to investments in fertilizer N. Although this approach has proven useful for demonstrating the concept of diminishing returns, it is nothing more than a black-box approach, which offers limited information for deriving N recommendations in another year or at another site. 1 IFDC/CIP, Lima, Peru. 2 INIA, Cajamarca, Peru. 3 INIAP, Quito, Ecuador. 4 CIP, Lima, Peru. Nitrogen is a dynamic and mobile nutrient, hence its effect on crop production is rarely the same from year to year. At best, equations from variable N-rate experiments describe only historical relations in the data. As such they offer little insight into the processes that must be understood to manage N inputs appropriately. More informative approaches for making N recommendations have followed from a better understanding of the processes that affect both crop N demand and soil N supply. That has led to N recommendations based on the simple principle that the amount of additional N to be applied can be estimated by knowing the crop N requirement and the potential soil N supply. The internal N needed to obtain a specific yield sets the crop N requirement. Soil N supply is set by the properties of the soil, particularly the soil organic matter (SOM) content and other properties that affect the extent and rate of microbial decomposition. Both also depend on the vagaries of weather. For high-yielding crops in most soils, the crop N requirement will exceed the soil N supply. It is the difference between these two that approximates the amount of supplemental N needed by a growing crop. Generally, the amount of N fertilizer to apply is calculated by estimating the soil N supply, a target yield level, and an expected fertilizer efficiency; this last because not all applied N is normally recovered by the crop (Dahnke and Johnson, 1990). To estimate each of these components, measurements of one or more of the following variables may be needed: whole plant N content required to reach a target yield, mineral N released during soil incubation, total N content of the soil, CIP Program Report

2 mineral N present in the soil profile at planting, and leaf N concentration or chlorophyll content at a specific growth stage. The relation between any of these variables and the benefit of added N is usually defined in field calibration studies. Because of site-specific effects, such studies need to be conducted across a broad range of soil-crop-climate conditions. There is, however, a practical limit to the number of soil-crop-climate conditions that can be included in field studies. An additional tool now available for evaluating N management practices and for making N recommendations is one based on computer simulation. Using dynamic simulation techniques, scientists have been able to construct computer models capable of simulating many of the major processes that control N demand and supply. When assembled within the framework of comprehensive crop-growth models, their dynamic nature makes them valuable for exploring N management options across a theoretically unlimited number of cropping practices, soil types, and weather conditions. The purpose of this study is to illustrate how we are using a potato growth model to better understand N dynamics in potatobased cropping systems. We stress, however, that simulation models are not a substitute for standard testing and monitoring of the N status in soils and plants. Rather, they are a tool for extending the value of such measurements by facilitating a systematic analysis of how plant, soil, weather, and management factors interact to affect N dynamics. In this study, we use the SUBSTOR potato model (Ritchie et al., 1995) to systematically analyze some of these factors to demonstrate how the model can be used to gain insight into N dynamics on a site-specific basis. We will also show results from model testing in Andean conditions. The version of SUBSTOR used in this study is that released with the Decision Support System for Agrotechnology Transfer (DSSAT v3) (Jones et al., 1998). A Comprehensive Crop Growth Model The SUBSTOR model provides a useful tool for analyzing the quantitative effect that controlled factors (e.g., management), uncontrolled factors (e.g., weather), and site-specific soil properties have on principal components of N balance. The model is comprehensive in that it simulates the major processes associated with crop N demand and soil N supply (Table 1). The model provides a balanced approach in the level of detail used to describe soil and Table 1. Major processes that are simulated and environmental factors that affect those processes in the N submodel of SUBSTOR-Potato (DSSAT v3). Process simulated Crop N demand Growth Development Soil N supply Mineralization/immobilization Nitrification Denitrification - NO 3 leaching Urea hydrolysis Uptake Main factors influencing process Solar radiation, temperature Photoperiod, temperature Soil temperature, soil water, C:N ratio + Soil temperature, soil water, soil ph, NH 4 concentration Soil temperature, soil water, soil ph, soil C Drainage Soil temperature, soil water, soil ph, soil C Soil water, inorganic N, crop demand, root length density 382 Natural Resource Management in the Andes

3 plant processes. All of the processes listed in Table 1 are simulated using a daily time step. Because it is comprehensive and dynamic, the model can be used to evaluate, on a site-specific basis, many alternative management practices against the uncertainties of weather. Simulation results, when based on reliable input data, can then provide critical information for defining the best N management practices from both an economic and environmental perspective. We have used experimental data from Ecuador and Peru to first test the performance of SUBSTOR under Andean conditions (Figure 1). In Ecuador, Clavijo (1999) showed that the model accurately simulated the response of two cultivars to N fertilizer at two sites in Carchi. In Peru, Yauri (1997) showed that SUBSTOR realistically simulated the growth of two cultivars with and without irrigation at a site in Huancayo. A comparison of simulated and observed tuber yields from the Ecuador and Peru studies is shown in Figure 1. This limited testing shows that the model realistically simulated tuber yields that ranged from 16 to 56 t/ha due to differences in weather, soils, cultivars, and management. Further testing is underway as we continue to critically evaluate the performance of SUBSTOR and search for ways to improve it as both a research and management tool. Simulated tuber yield (t/ha) Observed tuber yield (t/ha) Figure 1. Relation between simulated and observed fresh weight of potato tubers for sites in Ecuador and Peru (dotted line is 1:1 line). Components of a Systematic Analysis A general approach for estimating the amount of N fertilizer (N f ) to apply to a crop may be based on the following calculation: N f = (N y - N s )/E f where N y is crop N demand (the internal plant N required to attain the expected yield), N s is the N supplied by the soil, and E f is the expected efficiency or fraction of applied N the crop is expected to recover. This approach provides a useful framework for examining the main factors to be considered when managing N inputs. Within this framework, the N balance in agricultural fields can be characterized by defining three major components: N y, N s, and N f. To optimize N management for a specific situation, quantitative estimates of each of these components are needed along with economic information. Crop N demand Crop N demand is the product of the expected yield and internal N requirement, which can be thought of as the minimum amount of plant N associated with maximum yield (Stanford and Legg, 1984). Although a growing crop may take up more than the minimum N needed, extra N (luxury consumption) does not usually result in any yield benefit. Therefore, to optimize N management and avoid its inefficient use, it is important to know the expected maximum yield and its associated internal N requirement. Maximum yield, however, is not a constant. For a single cultivar, maximum yield will vary from site to site and year to year due to the interaction of genetic traits with photoperiod, temperature, solar radiation, water, nutrient availability, and management. Many of these interactions can be captured in the SUBSTOR model. To illustrate, Figure 2 shows how tuber yields responded differently to applied N when the SUBSTOR model was run using different weather years for a site in Huancayo, Peru. The cultivar simulated was Yungay, which was planted in mid- CIP Program Report

4 Tuber fresh weight (t/ha) N applied (kg/ha) Figure 2. Simulated response of potato tuber fresh weight to applied N using daily weather data from or for a site in Huancayo, Peru. November and harvested in early April. Whereas fresh yields reached a maximum of 55 t/ha using daily weather from , the same amount of N in produced a maximum yield of only 26 t/ha. The difference in N response between seasons was attributed mostly to a less favorable distribution of rainfall during The simulation showed the plants suffered a significant water deficit during tuber bulking. Although yield levels varied for the two seasons, the internal N requirement remained constant at about 16 g N/kg of total dry matter (DM) (tops plus tubers), which is within the range reported by Vos (1995) for several field experiments. The simulated removal of N in fresh tubers was about 2.8 kg N/t fresh weight in each season, also within the range reported for real data by Harris (1992). Simulation can be used as well to estimate the time course or N uptake pattern of a growing crop. With output provided on a daily basis, the model makes it possible to define the period of greatest N demand. Such information can then be used to examine how management might improve the synchrony between crop N demand and N supply. For example, the economic and yield impact of split applications of N fertilizer could be easily studied in simulations designed to vary the amount and timing of N applications. Soil N supply Nitrogen supplied by the soil comes mostly from two sources: 1) mineralization of soil organic N during the growing season, and 2) mineral N initially present in the soil profile at planting. Both sources should be considered when estimating the amount of supplemental N needed by a growing crop. However, the importance of initial mineral N tends to diminish in high rainfall environments where significant leaching can occur. Since the mineralization of soil organic N is a biological process, the amount of N made available depends primarily on the level of microbial activity and the amount of carbon (C) substrate. For most mineral soils, the C:N ratio in SOM is fairly constant at 10. That ratio favors a net release of N to available mineral forms unless OM is added with C:N ratios greater than 25. At a C:N >25, there is usually a loss of plant available N as it becomes tied up through net immobilization. The amount of mineral N present in the soil profile at planting (initial mineral N) often has a substantial impact on the need for supplemental N, particularly in less humid environments. Initial mineral N usually varies across sites and years, with the amount largely determined by management and growth of the previous crop and the residual N left from earlier applications. If rainfall is not excessive, much of the initial mineral N can remain available to a crop throughout the growing season. The process descriptions in SUBSTOR enable capturing the interaction of these sources of available plant N with crop N demand for innumerable combinations of soil type, weather, cultivar, and management. For example, simulation could be used to examine how soil N supply might vary across years for different quantities of SOM and mineral N present at planting. Moreover, a simulation study like the one 384 Natural Resource Management in the Andes

5 used to derive Figure 2 would obtain somewhat different responses if SOM or initial mineral N inputs were varied. Sources of supplemental N In SUBSTOR, supplemental N can be added as a mineral fertilizer or as a plant residue such as green manure (GM). Algorithms dealing with the transformation of animal manure have not yet been included, although efforts to do so are underway. Nitrogen management practices that can be examined with the model include the effect of varying N rates, time of application, placement depth, and fertilizer source. These practices can be studied for their effect not only on tuber yield, but also on nitrate leaching or economic returns. Such studies can be simulated for a single growing season, or across many seasons, to quantify the impact of weather variability and its associated risk. A simulation example that illustrates how the model might be used to estimate the yield benefit of alternative N management options is shown in Figure 3. In this example, SUBSTOR was run for 19 seasons using daily weather data from Huancayo. Cumulative probability N GM 250 N Tuber fresh weight (t/ha) Figure 3. The cumulative probability distributions for potato tuber fresh weight resulting from simulations made across 19 years of weather from Huancayo, Peru. The N management treatments were no applied N (0 N), green manure incorporated (GM), or urea applied at 250 kg N/ha (250 N). Once again, the cultivar simulated each season was Yungay, planted in mid- November and harvested in early April. The N management options were no external source of applied N (0 N), a legume green manure (GM) (4 t DM/ha, N content 2.5%) incorporated just before planting, and 250 kg N/ha applied as urea in two, equal split applications. The simulated tuber yields for the 19 seasons are plotted in Figure 3 as cumulative probability distributions. They clearly show the superiority of the urea treatment in increasing yield, but they also demonstrate the risk of low yields despite heavy N application. For example, because of poor rainfall distribution and subsequent drought stress, the probability of obtaining essentially no response to applied N would be about 15%, or about 1 yr in 7. But the probability of obtaining a significant response to N is much greater, with yields of t/ha expected about 50% of the time. With this type of information provided on a site-specific basis, decisions regarding N management can be made based on an improved awareness of both the potential and limitations of a given environment. To make a simulation even more useful, model inputs and yields could be combined with actual price and cost data to obtain similar distributions for net returns. That would better define the economic risk, which is of more interest to a farmer. Conclusions Nitrogen management can be improved through the insight provided by comprehensive crop simulation models such as SUBSTOR. As a continually evolving tool, such models have the potential to help researchers and farmers better understand how soil, crop, weather, and management factors interact to affect crop N demand, soil N supply, and fertilizer use efficiency on a site-specific basis. Any number of management scenarios can be examined and compared for their impact, not only on economic returns but also on the potential CIP Program Report

6 for excessive leaching of nitrates. Economic and environmental (nitrate leaching) risks due to uncertain weather can also be quantified. References Cited Clavijo, N Validación del modelo de simulación del sistema DSSAT en el cultivo de papa (Solanum tuberosum L.) en las condiciones del Cantón Montufar, provincia del Carchi. Eng. Thesis. Escuela Superior Politécnica de Chimborazo, Riobamba, Ecuador. 85 p. Dahnke, W.C. and G.V. Johnson Testing soils for available nitrogen. In: Westerman, R.L. (ed.). Soil testing and plant analysis, Third Edition. SSSA, Madison, WI, USA. p Harris, P.M Mineral nutrition. In: Harris, P. (ed.). The potato crop. Chapman & Hall, London, UK. p Jones, J.W., G. Tsuji, G. Hoogenboom, L.A. Hunt, P.K. Thornton, P.W. Wilkens, D.T. Imamura, W.T. Bowen, and U. Singh Decision support system for agrotechnology transfer: DSSAT v3. In: Tsuji, G.Y., G. Hoogenboom, and P.K.Thornton (eds.). Understanding options for agricultural production. Kluwer Academic Publishers, Dordrecht, Netherlands. Ritchie, J.T., T.S. Griffin, and B.S. Johnson SUBSTOR: functional model of potato growth, development and yield. In: Kabat, P., et al. (eds.). Modelling and parameterization of the soil-plantatmosphere system. Wageningen Pers, Wageningen, Netherlands. p Stanford, G. and J.O. Legg Nitrogen and yield potential. In: Hauck, R.D. (ed.). Nitrogen in crop production. ASA-CSSA- SSSA, Madison, WI, USA. p Vos, J Nitrogen and the growth of potato crops. In: Haverkort, A.J. and D.K.L. MacKerron (eds.). Potato ecology and modelling of crops under conditions limiting growth. Kluwer Academic Publishers, Wageningen, Netherlands. p Yauri, H Validación de un modelo para simular el crecimiento del cultivo de papa. Eng. Thesis. Universidad Nacional Agraria, La Molina. Perú. 98 p. 386 Natural Resource Management in the Andes